Michael R. Clement

Live-fly, large-scale field experimentation for large numbers of fixed-wing UAVs

In this paper, we present extensive advances in live-fly field experimentation capabilities of large numbers of fixed-wing aerial robots, and highlight both the enabling technologies as well as the challenges addressed in such largescale flight operations. We showcase results from recent field tests, including the autonomous launch, flight, and landing of 50 UAVs, which illuminate numerous operational lessons learned and generate rich multi-UAV datasets. We detail the design and open architecture of the testbed, which intentionally leverages low-cost and open-source components, aimed at promoting continued advances and alignment of multi-robot systems research and practice.

Multi-UAV software systems and simulation architecture

As unmanned aerial systems (UAS) continue to increasingly require greater integration of sophisticated software systems, developing and utilizing best practices and principles of formal software systems engineering can enhance and ensure the safety, reliability, and performance of these systems. This paper highlights the detailed implementation of a number of such tools, including agile software development methods such as automated software testing, and enhanced simulation-in-the-loop testing for multi-UAS virtual and live-fly capabilities. Significant and tangible benefit to active field experimentation is demonstrated through description of these integrated approaches, impacting ongoing efforts in multi-UAS research, testing, and assessment practices.

Exploring Network-Centric Information Architectures for Unmanned Systems Control and Data Dissemination

Abstract : The exploration and development of an information architecture for networked unmanned systems is described. The unmanned systems discussed utilize standard components for guidance and navigation, coupled with additional computing devices for interfacing with a network. These platforms in turn communicate with a broader network of devices, applications, and users via a variety of wireless network links. Networking a platform that is traditionally operated via serial control links and analog sensor downlinks provides two distinct advantages: (i) high-level control, or tasking, of the platform is easily extended from the single operator to any authorized user on the network; and (ii) sensor data and status information may be disseminated rapidly across the network to all interested recipients. The architecture developed through this exploration is applied in a prototype UAV which is utilized as both a high-resolution imaging platform and a wireless network relay. Testing and evaluation of the architecture occurs on an ongoing, quarterly basis through a cooperative field experiment program run by U.S. Special Operations Command and the Naval Postgraduate School.

Development, System Integration and Flight Testing of a High-Resolution Imaging System for Small UAS

The development and flight testing of a high-resolution imaging system for small unmanned aircraft systems is described. The system utilizes an off-the-shelf camera coupled to an onboard computer and a wireless network to provide very high quality imagery from a very low cost platform with a simple web-based tasking and data retrieval interface. The project incorporates three primary developments: (i) control over a tactical wireless ad-hoc network, (ii) an advanced path-following flight control algorithm that couples the flight and camera control, and (iii) a remote control capability for the sensor. The camera is a dual use sensor, providing full frame/rate video as well as 12MP digital still images, and a gimbal provides a limited pointing capability. The path-following flight control system allows an untrained operator to scribble a path on a digital map, which becomes the ground-track for the sensor. The aircraft autonomously determines the optimal flight trajectory to keep the sensor footprint on this track. A robust wireless mesh network integrates the aircraft with the tactical network, offering control of autopilot and sensor functions from any other node on the network. The complete system is evaluated in the joint Cooperative Field Experiments conducted quarterly by U.S. Special Operations Command and the Naval Postgraduate School, where operators put the system to use in realistic scenarios.

Multi-swarm Infrastructure for Swarm Versus Swarm Experimentation

This paper builds on previous Naval Postgraduate School success with large, autonomous swarms of fixed-wing unmanned aerial vehicles (UAV) to provide infrastructure for the simultaneous operation of multiple swarms. Developed in support of an event fostering swarm capability development through competition, the online referee, or Arbiter, monitors and evaluates multiple independent but interacting swarms. This Arbiter provides sensor modeling for both swarms, evaluation of inter-swarm interaction, scoring and enforcement of competition rules, and graphical display of game status. Arbiter capability is demonstrated through live-fly experiments and software-in-the-loop simulation. The Arbiter is also used to evaluate swarm behaviors that are developed for air-to-air pursuit of an opposing swarm with results provided in this paper.

Consensus-based data sharing for large-scale aerial swarm coordination in lossy communications environments

Increasing unmanned aerial vehicle (UAV) capabilities and decreasing costs have facilitated growing interest in the development of large, multi-UAV systems, or swarms. The constrained communications environments in which these swarms operate, however, have limited the development of behaviors that require a high degree of deliberative coordination. This work presents two algorithms that use a consensus-algorithm approach to reliably exchange information throughout large swarms as a means of facilitating swarm behavior coordination. Results from experiments conducted in simulation and live-fly exercises are presented and discussed.

Programmable Diagnostic Network Measurement with Localization and Traffic Observation

As networks become increasingly complex and pervasive, understanding and evaluating their running behavior and diagnosing configuration problems becomes more challenging and yet more important. This motivates a need to craft new diagnostic measurements suited to particular network environments and applications. However, once measurement protocols are in place on network devices it becomes difficult to modify them to new needs. Others have explored programmatic approaches that allow executing custom code at otherwise “unintelligent” network devices in order to provide configuration management and define new services. This approach can also be used to make meta-level observations from within a running network. We introduce a programmatic approach to diagnostic network measurement that offers such observation. It gives users a language in which to express measurements succinctly and an execution platform that enables network observation and localization of measurement. The design of the language and its platform are sketched with an example application.